Two Groups of Measured Findings
Analyze organizes into two labeled groups of findings, plus a separately labeled subjective read at the bottom of the list. There's no overview screen to open first and no numbered sequence to work through — each is a flat list of facets, and clicking Analyze takes you straight to the first one.
Every facet row shows how much of your manuscript it examined and, when it has something to say, how many findings are open. Those two numbers already tell you where to look — there's no dashboard to open first.
How a Finding Works
Each facet keeps two things apart: coverage — what it looked at — and findings — what it surfaced. A facet that reads "reviewed" is telling you it examined your manuscript and found nothing to flag, not that your manuscript is good. A facet that hasn't run yet says so outright, rather than showing a bare zero that could be mistaken for a clean pass.
A finding itself has a plain claim on the card face — no hedging, no drill-down required to understand it — and a "See details" link to the fuller reasoning and evidence behind it. Nothing in Analyze changes your manuscript or writes anything to your project on its own.
For each finding you decide: Add To My Actions puts it on your own revision checklist; Set Aside clears it for now without grading it, so a later read won't force you to re-triage it; and, from the finding's own detail view, Decline records that the read was wrong. None of the three touches your prose — the edit that follows is yours.
Record Gaps and Beat Gaps
Below the two measured groups sit the pages of what the analysis found that your records don't have yet. Record Gaps — Character Gaps, Location Gaps, and Timeline Gaps — compares the characters, places, and dates your prose contains against your own records and reports the difference; it is the only place in Analyze that writes to your project, and only when you say so. Beat Gaps lists the beats analysis detected in your scenes that are missing from the Structure area, with Add, or Add all for a chapter, to bring any of them into your structure as your own.
Editorial Read: One Reader's Opinion
At the very bottom of Analyze's list, behind a visible divider, sits the Editorial Read — the model's own developmental take on how your book is built and how it moves, written in prose rather than as a list of findings. It never shares a screen with the measured facets above it, and it's never presented as a finding: hand the same manuscript to two human editors, or two AI platforms, and you'll get two different reads back.
Weigh it the way you'd weigh notes from an editor you hired — worth considering, never a verdict to act on. At the foot of the read, "This resonates" and "I disagree" let you say whether it was useful; your answer persists across sessions. If you'd rather see only measured, denominatored findings, turn it off in Settings › AI Configuration › AI features — it disappears from Analyze's list entirely and everything else keeps working exactly as before.
Character Presence Has Moved
Character presence and mention density now live under Structure → Characters, on the Appearances tab, alongside your own scene-tag map — everything about a character in one place rather than split between Structure and Analyze. See the Character Presence & Arcs guide for how to read those maps.